Hi Everyone,

I am a bit busy with the API 650 Appx P at the moment and I see there is conflict on Upper Limit (u) in Table P-2 for equation numbers 11-29 and Figures P-8A, P-8B.

By looking at Fig P-8A and P-8B the upper limit (u ) should be the following:

For Equation No 11

Hi Everyone,

I beg your pardon about the unfinished thread above.

I am a bit busy with the API 650 Appx P-2009 at the moment and I see (my feeling) there is conflict on Upper Limit (u) in Table P-2 for equation numbers 11-29 and Figures P-8A, P-8B.

By looking at Fig P-8A, P-8B, P-8C and P-8D the upper limit (u ) should be the following:

For Equation No 11, 0.3
For Equation No 12, 1.0
For Equation No 13, 2.0
For Equation No 14, 5.0
For Equation No 15, 0.5
For Equation No 16, 1.0
For Equation No 17, 2.0
For Equation No 18, 3.0
For Equation No 19, 5.0
For Equation No 20, 1.0
For Equation No 21, 2.0
For Equation No 22, 3.0
For Equation No 23, 5.0
For Equation No 24, 5.0
For Equation No 25, 1.0
For Equation No 26, 2.0
For Equation No 27, 4.0
For Equation No 28, 5.0
For Equation No 29, 5.0

However Table P-2 gives the sequence for upper limit values upside down for each Figure.

Has anyone come across with the same conflict? What do you think which information is to be correct in this code? Or Am I missing something?

Can you please comment on this issue?

Thanks in advance and kind regards,

Ibrahim Demir

Ibrahim,

Section P.3 has been deleted from API-650, as of 11th Ed Addendum 2. (Too many problems I suspect.)

Richard,

Thanks for the information, I did not check the latest version of the code to see that. It seems that I need to obtain the paper referred.

However, it is a bit shame for the API code committee not to maintain the section adequately within the code. This is not the first time I came accross this. What was the reason they introduced this section into the code at the forst place?

I should not complain too much. At least they refer the original paper.

Kind regards,

Ibrahim Demir

Richard,

I beg your pardon. API 650 does not refer anything for the deleted section under P3, the reference "Analysis of Loads for Nozzles in API 650 Tanks" is part of the sample problem in section P2.
However the following note in P1 still remains.

"This appendix presents two different procedures to determine external loads on tank shells. Section P.2 establishes limit loads and
P.3 is based on allowable stresses. This appendix is based on H. D. Billimoria and J. Hagstrom’s “Stiffness Coefficients and
Allowable Loads for Nozzles in Flat Bottom Storage Tanks” and H. D. Billimoria and K. K. Tam’s “Experimental Investigation
of Stiffness Coefficients and Allowable Loads for a Nozzle in a Flat Bottom Storage Tank.”".

I am not sure that getting these two papers will be useful on the issue.

Kind regards,

Ibrahim Demir

Ibrahim,

API 650, 1.1.20 says "Appendix P provides minimum recommendations for design of shell openings that conform to Table 3-6 that are subject to external piping loads. An alternative or supplemental design may be agreed upon by the purchaser or manufacturer."

So under agreement you may use anyway method "P3 corrected", or better said, WRC Bulletin 297 method adapted for tank calculation.
As a part of this agreement, you may consider WRC 297 in order to calculate the dimensionless factors ni, mi and fi. Otherwise you cannot prove API650/P3 mistakes are in graphs or they are in equations.

The problem of this method is that the mistakes are coming up in the original ASME paper which is "Analysis of Loads for Nozzles in API 650 Tanks, M. Lengsfeld, K.L. Bardia, J. Taagepera, K. Hathaitham, D.G. LaBounty, M.C. Lengsfeld. Paper PVP-Vol 430, ASME, New York, 2001"
I cannot blame API 650 committees in this case.

Regards

Mariog,

Thanks for your contribution to the issue.
I do not know how you get this information. I went through the history side of API 650, and could not find any explonation at all. I would expect a short description in the history or in the code why the section was deleted at least. It seems that either it does not exist or I overlooked and could not see it.

I did not say anything about the formulation and others. I have just concentrated on the upper limits of "u" values for 4 figures only as I mentioned above. In case the original paper contains mistakes as you mentioned there is no hope to work on the nozzle strength calculation. Can I ask how did you obtain this information? I may have to present the same information to others within the company.

In the tanks we are working on there are some nozzle at centreline distance to bases 1500-1600 mm. This creates problem. The only solution might be to use WRC 297 or FEA. We are doing the piping stress analyses, therefore FEA is not an option for us for this stage of the job.

So, we may be forced to use WRC 297 option of Caesar II. I was wondering if the CAESAR II WRC 297 is still valid option for the solution?

Kind regards,

Ibrahim Demir

In 2005 I was enthusiastic about the "new" method based on WRC, hoping the conservative Billimoria's approach would be history. I've input the equations in MathCad and...I've seen the inconsistency between equations and graphs.

I bought on-line the original article and I was totally disappointed. I realized it wasn't API 650 mistake, the original article is full of mistakes.

Moreover, the original article does not offer any new clarification, so I would say money spent for nothing.

In fact, the procedure explained in article is identical rewritten in API 650 - see for example 11th edn, addendum 1, P.3.4 STRESS FACTORS DUE TO FR, MC, ML
Total stress has three components: bending, membrane, and shear. The stress factors for calculation of stresses in the shell at the nozzle junction have been presented for the stress equations. Moment loads include circumferential and longitudinal moments.
σr = (FR/t^2)(nr ± 6mr) (Radial Load) (1)
σθ = (FR/t^2)(nθ ± 6mθ) (Radial Load) (2)
σr = [Mi/(dt^2)](nr ± 6mr) (Moment Load) (3)
σθ = [Mi/(dt^2)](nθ ± 6mθ) (Moment Load) (4)
The dimensionless factors ni and mi were obtained from WRC Bulletin 297, Figures 3 through 58 for various nozzle sizes, tank diameters and thicknesses. The absolute maximum of these values as used in the stress Equations 1 through 4 have been combined into the dimensionless stress factors:
fi = (ni ± 6mi)
The factors were plotted after transformation into mathematical equivalents in Figures P-8 through P-10.

Well, the only problem is the math is not "equivalent" with graphs. Reading article, at least you can understand that the authors were confident in their equations; they said "due to space limitations, not all curves are included in this paper.
Additional curves can be produced using the mathematical
equations from Table 1."

I suspect the mistakes were made by a draftsman that destroyed the graphs of a good article. Of course at this level you expect authors (at that time Fluor Daniel's specialists M. Lengsfeld, K.L. Bardia, J. Taagepera, K. Hathaitham, D.G. LaBounty, and CCI's M.C. Lengsfeld) might have checked out the final article.

For me it was comical to see six years of API efforts to correct this problem, but I cannot blame API committee more than, in the end, it seems that they gave up. Probably it was a consistent negative feedback from industry. However-in my opinion- there is not any error method there, there are stupid uncorrected graph mistakes in that article.

About your question: in my opinion, under an agreement described by API 650 in 1.1.20, you may use any WRC procedure that offers you "ni" and "mi" and after you would calculate fi = (ni ± 6mi), exactly as article intended. You would apply Stress reduction factors as article/method P3 considers and you need to apply the allowable stress as they considered there. In the end you will be able to document the results; however don't forget to obtain the API 1.1.20 agreement! I'm not so optimistic; since you are not tank manufacturer/ purchaser, you cannot force them to follow your procedure...

Regards.

Mariog,

Thanks for the input. I have just received the copy of the paper. For a short look, I see that the equations in P-3 and in the paper are the same in terms of upper limit "u".

Unfortunately, the paper graphs are limited to d/t = 50. By comparing these graphs I can see that the curves in Fig P-8C and P-8G in API 650 were marked wrong if we assume the paper is correct. I do not want to generalize this for the other curves above because thay are not in the paper.

I guess the authors can correct the mistakes either in the code or in the paper by looking at the findings of their FEA analyses. So I would expect the Appx P-3 will come back in the new version or amendement.

Thanks again.

Ibrahim Demir